BACKGROUND

Cardiovascular disease and chronic kidney disease (CKD) are two major complications of Type 2 diabetes (T2D) and the leading cause of mortality worldwide(Kalantar-Zadeh, et al., 2021; Roth, et al., 2020). Observational studies have found that lipid traits were associated with the risk of atherosclerotic cardiovascular disease (ASCVD) and CKD in T2D(Colhoun and Marcovecchio, 2018; Newman, et al., 2017). However, the causal nature of the observed associations are unclear. We aimed to determine the causal associations between circulating lipid traits and ASCVD and CKD in T2D using both linear and nonlinear Mendelian randomization (MR) methods.

METHODS

We prospectively analyzed 9,067 White British participants with T2D at baseline from the UK Biobank. A total of 6 lipid traits including low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglycerides (TG), apolipoprotein A (ApoA), apolipoprotein B (ApoB) and lipoprotein A (Lp(a)) were used as exposures. The outcomes of this study were ASCVD and CKD, which were defined based on the ICD-10 codes. Follow-up time was defined as the period from baseline to the date of the first clinical endpoint, or the censored date of 12 November 2021, which came first. We extracted genome-wide significant SNPs associated with lipid traits from previous published GWAS, and excluded those with minor allele frequency<0.01, Hardy-Weinberg equilibrium P<1×10^-5 and LD r² >0.1 within 1Mb, as well as those associated with confounders of smoking status, alcohol consumption, Townsend deprivation index, physical activity, blood pressure, body mass index and HbA1c. Finally, we generated the weighted genetic risk scores (wGRSs), which were employed as instrumental variables (IVs), using 65, 69, 43, 269, 161 and 4 SNPs for LDL-C, HDL-C, TG, ApoA, ApoB and Lp(a), respectively. A two-stage method and nonlinear MR with a piecewise linear method (Staley and Burgess, 2017) were used to assess the potential linear and nonlinear causality relationship between each lipid trait and the outcomes .

RESULTS

Among 7,067 individuals without ASCVD at baseline (mean age 60.43±6.63[SD] years; 39.31% female), 1,898 (26.86%) progressed to ASCVD during a median follow-up period of 12.62 (IQR, 11.80-13.46) years. There were 1,611 (18.72%) out of 8,607 participants without CKD at baseline who progressed to CKD during the periods. The F-statistic for wGRS were all >10. In linear MR, higer genetically predicted LDL-C was associated with a higher risk of ASCVD (HR, 1.40; 95% CI, 1.12-1.76; P=0.004; Table 1), but not with CKD (P=0.162; Table 1).  One SD increase in genetically predicted ApoB was associated with 1.30- and 1.43-fold risk of ASCVD (95% CI, 1.04-1.64; P=0.023; Table 1) and CKD (95% CI, 1.12-1.83; P=0.005; Table 1) in individuals with T2D, respectively. There was no evidence for causal relationships of HDL-C, TG, ApoA and Lp(a) with ASCVD and CKD. Nonlinear MR provided no evidence of nonlinearity for the causal associations of lipid traits with ASCVD (Figure 1) and CKD (Figure 2) in individuals with T2D.

CONCLUSIONS

Lower ApoB concentration is causally associated with lower risks of both ASCVD and CKD in T2D. The findings suggest that ApoB may be a better therapeutic target in prevention of cardiovascular and kidney diseases in T2D, when compared with LDL-C.

type 2 diabetes;,chronic kidney disease,Mendelian randomization,lipid profile,atherosclerotic cardiovascular disease